Archives of the TeradataForum

Message Posted: Wed, 07 Oct 2009 @ 20:08:02 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

Eric,

This sounds interesting. No, I don't know anyone that has done something like this. You said in your response:

I don't think a workload throttle with a limit of zero would do what you want. That would result in a growing delay queue. And I don't know of any way you can automatically or programmatically abort from the delay queue, although I like your thinking. Neither can you tag which queries were delayed once they start running in the workload and set up an immediate exception on them and only them.

Here's the only options I see:

1.) If you know all queries that classify to the workload are ones you want to reject, then you could try to make the exception on that workload so tight that each query can only run for a small amount of time before being aborted. That's not very clean, you waste resources, including AMP worker tasks,locks, etc. But the big problem is defining a tight enough exception and catching the exception early enough. I've never seen anyone set the exception interval much below 10 seconds, which means each query will run for 5 seconds on average. That assumes you can figure out how to reach an exception almost immediately.

2.) You might be able to use a queryband. You can associate a queryband name/value pair to a filter in Teradata 12, and filters are the easiest vehicle when you want a clean, immediate rejection. But as I said earlier, you have to have a mechanism in place for setting the queryband before the query is submitted.

Thanks, -Carrie

	<Prev	Next>		<<First	<Prev	Next>	Last>>